Luuk

Hetzelfde*

De absoluut 100% garantie wanneer iemand je rondleidt door een kantoortuin en dan zegt “haha ja het zijn flexplekken maar je ziet toch altijd dat mensen op dezelfde plekje gaan zitten” ja uhh misschien werkt dit concept ook gewoon helemaal niet dan?

Lord of the rings ass building

That’s the other end of it. Which of course is also bad. I’m coming at this with mostly machine learning and psychology papers where there is no narrative at all and at worst it’s impossible to replicate the study because they just don’t say what they did and at best just left out a lot things

Maybe I laid it on a little thick. I was not coached in the sense that I was literally told to be ‘more convoluted’, but definitely suspiciously vague advice that particular sections are ‘unnecessary’ or ‘not academic enough’ or being told to cite some paper that’s not particularly relevant

Do these ring true?
- math seemingly plucked from the mind of god, in reality the result of trying out ideas that didn’t work
- experiments that only give you a vague idea about the actual task
- difficult to read paper, but perfectly clear talk where the researcher communicates like a human

These are all habits I’ve been actively coached on at some point, but actively hurt the readability of your paper. They just serve to meta communicate “hey this work is publication worthy, look at how complex it is!”

I wouldn’t want to “simplify” it so much, rather:
- not write in convoluted prose
- unnecessary/unexplained math
- have no (fun) examples
- leave out parts of the process or things that didn’t work but were really important in building the intuition
- cite just to have a big bibliography
…

If I were to go back into academia I would aspire to write two versions of the research I’d do: the fun one written in my style for actually helping the reader understand the work and the fake one with all the performative bullshit and obfuscation to get through peer review

So true. The awkwardness when someone asks me “so what do you do” and I have to say I work with AI and also immediately hedge it that I don’t like the culture around it.

Je zou kunnen denken dat Steely Dan top2000-coded is maar dat is niet zo (slechts 1 nummer op 1890). Ik denk eigenlijk dat het té muzikaal is voor de gemiddelde Nederlandse boomer

But how can I be sure I’m practicing minimalism if my products don’t have any å

Als ik nog een keer “er is al belasting is betaald over dat bedrag” lees of hoor word ik de joker

What would be the gold data? Do you even need? I can imagine an Evolutionary algorithm would converge to something very similar to the objective of BPE

Beautiful pragmatics going on a work: Our mandatory “inclusion” e-learning mentions you shouldn’t comment on a colleague’s appearance (also not positively), and now I just overheard someone say “it’s a shame I’m not allowed to compliment your outfit”

Het yuppen equivalent van de Sovjetflat

Evolutionary Psychology ❌🤢

Cultural Evolution ✅😃

Rutte als NAVO SG over defensieuitgaven van Nederland be like “wie de fuck heeft er daar aan de knoppen gezeten de afgelopen 10 jaar”

What happened that we’re all collectively giving horses names like “Nice Café”

Exactly man! I’m telling you… a new party… yup in a FPTP voting system… I swear he’s not kidding!… so anyway I tell Elon about Duverger’s Law

I for one I am really glad they’re adding new loading animations instead of fixing the obvious memory leak in the Reddit app. It bricks my phone for like 2 minutes forcing me to stop doomscrolling!

I tell everyone I quit my PhD due to terrible supervision conditions, but secretly it was finding out about EXH that made me so alienated from linguistics I couldn’t bear it

Been a while for me, but conjuring up new covert grammatical operators alchemy style to deal with clearly non-truth conditional phenomena was really hot and maybe still is.

AFAIK it’s mostly linguists w/o very rigorous formal training who don’t realize they’re working within a framework with limits.

I think mate is slowly making its move on NL. I see it in more and more cafes now

uv is definitely a step in the right direction

If it’s any consolation it’s just Python and the problem is not limited to conda but every single package manager does this shit.

I’m not ready for 35 degrees man, I don’t know how to survive that

We were lucky enough to have our YouGov/The Economist poll in field when the US bombed Iranian nuclear facilities. You can see the partisan realignment in real time

A colleague of mine calls consultants ‘slide bakers’ and I think this is a term we should all adopt